计算机与现代化

• 数据挖掘 • 上一篇    下一篇

  基于LSTM的临床血液需求预测方法

  

  1. (江西省血液中心,江西南昌330052) 
  • 收稿日期:2017-10-24 出版日期:2018-06-13 发布日期:2018-06-13
  • 作者简介: 郑亚鹏(1978-),男,江西新干人,江西省血液中心工程师,学士,研究方向:网络安全管理,数据分析; 樊璐(1982-),女,吉林榆树人,主管技师,硕士,研究方向:输血检验。

 A LSTM-based Forecast Method for Clinical Blood Demand

  1.  (Jiangxi Province Blood Center, Nanchang 330052, China)
  • Received:2017-10-24 Online:2018-06-13 Published:2018-06-13

摘要: 为了帮助血液中心制定合理的采供血计划,本文根据江西省血液中心收集的2005-2016年临床用血数据,分析临床用血总量和各种血液成分用量的变化趋势,并基于LSTM神经网络预测2016年各月的临床血液需求量。实验结果表明,相较于传统的ARIMA时间序列模型,本文建立的LSTM神经网络能够有效预测临床用血需求量的变化,得到较为准确的预测结果。

关键词:  , 临床用血, 需求预测, LSTM网络

Abstract: In order to help the blood center make a reasonable plan, this paper analyzes the total amount of clinical blood and the changing trend of various blood components according to the clinical blood data from 2005 year to 2016 year collected by Jiangxi Province Blood Center, and predicts the clinical blood demand in every month of 2016 year based on the LSTM network model. Experimental result shows that, compared with the traditional ARIMA model, the LSTM network established in this paper can effectively predict the change of clinical blood demand and obtain more accurate prediction results.

Key words:  clinical blood, demand forecast, LSTM network

中图分类号: